************************
** Training 2 do file **
************************
** may 26, 2021 ********
************************


cd "E:\My Drive\Heal STATA Trainings\Training 2\data"

use "Training 2 data", clear


****************************************
* Objective 1 Conditional Statements **
****************************************

** use if statement **
	sum earnings if male==1
	sum earnings if male==0


** use and statement **
* average hours worked for males that are business majors *
	sum hours if male==1 & major=="business"
	
** use or statement **
* average hours worked for those who are men or those who are married *
	sum hours if married==1 | male==1
	
	

** conditional tabulations **
	tab age if education>12

** conditional count **
	count if education >12

** conditional sum **
	sum earnings if major=="business" 
	sum earnings if major~="business" 
	
	
	
****************************************
* Objective 2 Manipulate the data **
****************************************


* generate new variable *
	gen lit=1
	
* generate new variable based on a function of other variables *
	gen age2=age^2
	sum age2 
	
	gen hours_year=hours*weeks
	sum hours_year
	
*replace variable*
*dummy variables*
	gen female=0
	replace female=1 if gender=="female"
	
* create missing data *
// those that work more than 80 hours per week are miscoded and should be missing values 
	replace hours=. if hours>80
	
* create a single race variable *
	gen race=" "
	replace race="white" if white==1
	replace race="black" if black==1
	replace race="hispanic" if hispanic==1
	replace race="other race" if other_race==1

	tab race